Learning and remembering spatial relations is an important human capacity. Many lines of evidence suggest that the hippocampus is essential for this form of learning. However, much less is known of how spatial relations are encoded in the hippocampus or of how the hippocampus changes as learning of spatial relations occurs. This application will study the substrate for spatial learning using a novel animal model: seed storage in birds. Several species of birds hide large numbers of seeds and are able to retrieve them reliably weeks or months later. Several studies suggest that the hippocampus is necessary for the learning--it is larger in species that store than in related ones that do not, and birds that store are unable to retrieve if the hippocampus is lesioned. Experiments are proposed that will investigate the anatomical basis for the enhanced volume in species that store. This will suggest which aspects of the hippocampus have been augmented to permit this exceptional spatial ability. Other experiments will determine whether hippocampal anatomy in birds that store changes seasonally in phase with seasonal variation in the incidence of storage. If this were found, it would emphasize the extent to which the structure of complex neural networks can be modified in adulthood to allow for (or to encode) enduring changes in behavioral expression. Another experiment will determine whether short-term practice at storing and retrieving results in structural changes in the hippocampus. This experiment is designed so that storage and retrieval use one hemisphere, permitting within-animal comparisons of the consequences of learning. Differences in anatomy found in this experiment could point to sites and mechanisms used in spatial learning. It will be especially interesting to see whether any such changes are general or are focussed, perhaps in regions of the hippocampus that are enhanced in the birds that show the behavior. Together, these experiments assess this form of spatial learning at several levels. They search for network features related to the behavior, for long term structural change in the system related to changes in the frequency of the behavior, and for short term changes related to use. Each of these issues can be related to human spatial learning, its substrate and its capacity for change. The studies in this proposal will generate data that will be used to design a more complete program for investigating the neural substrate of spatial learning in this system.